pytorch-ewc

pytorch-ewc copied to clipboard

When we train the current task, will we use the data of the previous task？ ewc need task A data to compute fisher info, when we train task B how...

xiaobingbuhuitou

log-likelihood of regression problem

2

As the code, we can get negative log-likelihood by F.log_softmax. How can I get the log-likelihood of regression problem??

tytcc

I found that performing forward-backward in a loop is much faster than using autograd.grad with retain_graph=True. Your current code is: ``` loglikelihood_grads = zip(*[autograd.grad( l, self.parameters(), retain_graph=(i < len(loglikelihoods)) )...

htt210

labels are shared for every task

2

Great implementations, many thanks. I'm new to the continuous learning and a little bit confused about its problem setting. I see that you generated different permuted MNIST data for every...

Yijunmaverick

Fisher update

1

I am trying to run EWC on my dataset with resnet50 model. While updating the fisher matrix using your function, My code says Cuda out of memory due to "log_liklihoods.append(output[:,...

Sharut

Bumps [torch](https://github.com/pytorch/pytorch) from 1.0.1.post2 to 2.2.0. Release notes Sourced from torch's releases. PyTorch 2.2: FlashAttention-v2, AOTInductor PyTorch 2.2 Release Notes Highlights Backwards Incompatible Changes Deprecations New Features Improvements Bug fixes...

dependabot[bot]